INDEX

Explanations

proper nouns

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "It" indicating a shift to a new subject or point in the text

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

hips

-0.68

 Polk

-0.66

 Dayton

-0.61

Tur

-0.61

ded

-0.59

 Pearce

-0.58

 ears

-0.58

Uni

-0.57

 household

-0.57

 gift

-0.57

POSITIVE LOGITS

self

1.24

zbollah

1.08

unes

1.08

ueller

1.06

chwitz

1.01

chy

0.97

zik

0.95

iner

0.93

anyahu

0.91

alm

0.91

Activations Density 0.265%

proper nouns

occurrences of the word "It" indicating a shift to a new subject or point in the text

No Comments

No Known Activations